Bishop Hill Worse than we thought
Dec 6, 2011 Climate: Surface Anthony Watts has republished Hausfather (Mosher) et al's poster for the AGU fall meeting in San Francisco. It covers the urban heat island effect and is based on a better determination of which stations are urban and which rural.
Urban stations are warming faster than rural stations on average across all urbanity proxies, cutoffs, and spatial resolutions examined, though the underlying data is noisy and there are many individual cases of urban cooling. Our estimate for the bias due to UHI in the land record is on the order of 0.03C per decade for urban stations. This result is consistent with both the expected sign of the effect and regional estimates covering the same time period (Zhou et al 2004) and differs from some recent work suggesting zero or negative UHI bias (Wickham et al, submitted).
Zero or negative UHI always struck me as a bit daft. This result seems intuitively much more sensible to me.
Reader Comments (35)
Zero or negative UHI always struck me as a bit daft. This result seems intuitively much more sensible to me.
Same as Willis Eschenbach,s test Does it Smell Right
Pity a few well paid and biased to the cause Scientists do not apply it before they publish. Lets face it the 1990 Jones et al China UHI has no effect has done his reputation no good.
If Steve Mosher should visit I would like to know his views whether this new work now confirms the work done by John Van Vliet (John V) along with Steve McIntyre and others at CA in 2007 that showed a bias due to siting of 0.35C per century. Steve Mosher was involved with this along with Kirsten Barnes, Kenneth Fritsch, Clayton B and Anthony Watts amongst others. Things often got contentious with people throwing toys out the pram but there was also an impressive level of cooperation at times.
http://climateaudit.org/2007/10/04/gridding-from-crn1-2/
http://climateaudit.org/2007/09/15/a-second-look-at-ushcn-classification/
Tim Lambert and John Cook subsequently used this analysis to say that adjustments made by GISS were somehow ok. However then the Menne paper came out which appeared to show no difference based on siting and which caused problems with the Lambert and Cook reference to John Vs work. I have always been troubled by the discrepancy between John V and Menne et al.
This may not be the scientific way but as a motorcyclist I can say on a cold night that the difference in temperature between country and town is a big one you can actually feel it !
Don't worry - global warming has been solved.
This from Dennis Bushnell, Chief Scientist, NASA, Langley.
Full talk here
The UHI effect is one of those things like the Mediaeval Warm Period (there _were_ vineyards in Leeds), that is so obvious and self evident I'm surprised that mention of a zero UHI effect isn't laughed out of court (/peer review).
In winter, during the weather reports a frost or freeze is _always_ predicted to be worse in rural areas.
Nial.
A bibliography of articles on Urban Heat Islands, through March 3, 2010:
http://www.urbanheatislands.com/bibliography
mat
During the winter of 2009-10 I regularly drove home (10 miles) from Edinburgh at about the same time once a week for about 13 weeks.
The temperature where I lived was regularly 3 to 4 degrees C lower than in the city centre.
As they say, the plural of anecdote is not data but I was struck by the consistency.
Its the increase in the UHI effect thats important not the effect itself.
As the population increases and the proportion of the population in urban areas increases the increase in UHI gives an increased Global temperature average if no counter adjustments are made.
The Jones et al 1990 China UHI effect has been cited numerous times to support low or nil adjustments for the increased UHI effect over the years.
I am sure that any conclusions are just as accurate as the underlying temperature data.
This study has the same problem as all previous studies. They are comparing Urban with Rural without in depth studies to see if that is a reasonable split. We have always ASSumed that larger cities will have larger UHI TREND than smaller and rural. Dr. Spencer's study showed this may not be the case. Additionally, since there is no way they can statistically separate the many influences on the TREND, what they have shown is NOT UHI, but, just overall trend made up of UHI, Solar, GHG's, etc.
As far as negative UHI, I see no issue with this conceptually. If we start with a fast growing city where tress, shrubs, grass, etc. has been replaced with home, businesses, factories, blacktop, cement, etc. we can easil see that there will be a positive UHI TREND over the development period. If the city stops growing, the trend would go to zero. If the city built parks, covered roof tops with gardens or reflective materials, trees and greenery in medians and sidewalks, improved insulation and energy use efficiency, we COULD see a NEGATIVE UHI TREND!! Don't know if this has happened, but, imagine the old west ghost towns with a good weather station. Would have shown a negative UHI for a while with less complex reasons.
It is also important to consider microclimate as well. The problem with meteological stations is that they are designed to measure weather, not climate. Even a small change in microclimate in a completely rural area could affect the temperature readings. Simple things, like a hedge or tree growing, or a wall being built. Think of walled gardens and open gardens - huge change in microclimate and yet the population of the area changes not one iota. How to identify such contamination of rural (and urban sites) but local change? Virtually impossible over long periods of time. And this on top of the general UHI caused by population growth, construction etc
I have some numbers somewhere on the change casued by a hedgeline growing 150 feet from a stevenson screen. I'll try and dig them out because they demonstrate how remarkably sensitive weather stations are to small changes to the local environment.
I'm afraid I can't take this work seriously unless the authors lose all their data and get all queeny and paranoid if asked to produce it.
Everyone knows that's how climate science is supposed to be done. Right, "Professor" Jones?
I live in an urban area - I work 6 miles away just outside a village. I can tell you that going by the thermometer on the speedo of my bicycle - the temperature is sometimes around 2 to 3 degrees lower by the time I get to work.
@ Mike Jackson
As they say, the plural of anecdote is not data
Ooh, I dunno:
http://www.independent.co.uk/environment/snowfalls-are-now-just-a-thing-of-the-past-724017.html
I wonder what the difference in seasonal, yearly and decadel UHI would be for stations on a peninsula that is a major tourist region and expands its concrete highways to except ever more volumes of traffic during the summer months and also erects huge monoliths to the wind Gods.One thing I wouldn't expect the change to be, is simple or linear.
The survey that Anthony Watts is carrying out in the USA at
www.surfacestations.org
shows some amazingly poor positioning of instruments that are included in the temperature records.
To anyone who has been involved in the correct exposure of meteorological instruments, some of the examples are a joke.
I remember in the 1960's been involved in the re-siting of an instrument enclosure which was deemed too close to a building. The thermometer screen should be twice the height of the building away from that building. The enclosure was moved quarter of mile into open ground. Within a few years you could see the step change in the records. Even if the instruments are in an open aspect (except being to close to that building) UHI type effects are experienced.
It is obvious that poorly sited instruments, close to buildings, close to tarmac, close to air-conditioning units, especially where these things have been introduced over the years, and even increasing traffic at airports where jet-wash contaminates an accurate measurement -are going to add a warming trend. To argue otherwise is not really credible.
"The thermometer screen should be twice the height of the building away from that building"
And anemometers need to be ten times the maximum dimension of the building away to be unaffected. Difficult in some situations (e.g oil rigs) I grant you...
NASA found UHI of 7C - 9C in 42 cities in the Northeast USA using satellites.
http://www.nasa.gov/topics/earth/features/heat-island-sprawl.html
Why can't Zeke and Mosher find it in the temperature record?
I hesitate to post a study conceived as a climate alarum by Ken Livingstone, but (ignoring his projections for 2050) it explains
'London’s UHI was first ‘discovered’ at the turn of the 19th century by Luke Howard, who is
widely known as the man who named the types of clouds. Over the course of 9 years he noted
an UHI effect of approximately 2oC (warming) during the night and -0.2oC (cooling) during the
day. By the middle of the 1960’s an average difference of 4-6oC in nocturnal temperature
between the central city of London and its surroundings was evident. More recently urban
climatologists have noted extreme UHI intensities in excess of 7oC. For example during the
August 2003 heat wave, the UHI intensity reached 9oC on occasions (Figure 3).'
http://static.london.gov.uk/mayor/environment/climate-change/docs/UHI_summary_report.pdf
kuhnkat writes:
"This study has the same problem as all previous studies. They are comparing Urban with Rural without in depth studies to see if that is a reasonable split. We have always ASSumed that larger cities will have larger UHI TREND than smaller and rural. Dr. Spencer's study showed this may not be the case. Additionally, since there is no way they can statistically separate the many influences on the TREND, what they have shown is NOT UHI, but, just overall trend made up of UHI, Solar, GHG's, etc."
Very well said. I would put it a bit differently. I would say that they need to do some serious research to determine the facts on the ground before they apply their statistical techniques. We need to know what changes in local environments are worthy of the name "urbanization" and we need to know the effects on local temperature readings of these changes. I know it is too much to ask these people for physical hypotheses based on empirical research, but something like that, something better than Briffa's standards.
As for their analyses of results, their goal should be to explicate for the non-statistician how it is that their statistical work does justice to the facts on the ground. If they are not willing to provide such an explication then their audience is statisticians.
Bruce,
Excellent question. Would Zeke and Mosher please answer it?
Bruce, Theo, Zeke and Mosher aren't looking for UHI itself; they are looking for its effect on trends. In effect they are looking for the change in UHI arising from urbanisation. The problem is that if you compare "urban" data with "rural" data it is not immediately obvious which sort of site will have urbanised more, especially when you include the microclimate effects.
As Andrew suggests, it is "obvious" that changes in UHI arising from urbanisation make a contribution to temperature trends. But it is surprisingly hard to detect the effects convincingly, and so to know what fraction of the trend can be explained away. Steve Mosher has discussed this question quite extensively.
Jonathan Jones,
I am interested in the facts on the ground. Should I stop reading statistical analyses? Will no one ever pony up and explicate how their statistical work respects the facts on the ground? Is that simply beyond folk who work in statistics?
I think that JJ expresses the issue very cogently...of course Mosher wants to make his points in a passive-aggressive way because he is a strange guy, and refuses to admit the points that people make....such as that UHI is embedded in the thermometer record...but maybe he means well...like you would say of a polite crackhead
Uncertainty even hits Richard Black of the BBC....
http://www.bbc.co.uk/news/science-environment-16049028
Rick
Nice to see Black expressing (or at least reporting) some doubt. It was hard to see the point of some of the stats (e.g. Egypt could be better off by 100%, or worse off by 206%) but I suppose, like targets, the MO has to provide figures, however useless.
I uprated a few of the more cogent comments, but when I tried to downvote one, I got the message:
"We're having some problems rating this comment at the moment. Sorry. We're doing our best to fix it."
Hmm...
"NASA found UHI of 7C - 9C in 42 cities in the Northeast USA using satellites."
I seem to recall BBD saying that UHI wasn't detected by satellites, as they were reading temperatures higher up...
Theo
If you want good stats, look at O'Donnell etal's rebuttal of Steig 09. There seem to be very few in global climate records.
The problem with Hausfather and Mosher seems to be that they persist in using the data in the climate records at face value. While this is better than subjecting it to arbitrary and poorly-documented adjustments, as GISS does, it leaves unaddressed the issue of the extent to which the category of "rural" stations includes those that were once truly rural and are now suburban. Even a small number of such stations will impart a significant warming trend to the overall data set, so simply subtracting the "rural" trend from the "urban' one and announcing the difference as UHI falls well short of quantifying the effect, however carefully they try to dice the urban v rural. It is the history of the rural stations - which with sprawl now fail to meet Zeke's strictest criteria but 30 or 40 years ago would have done - that holds the key to a more accurate assessment of UHI. He seems to be a very capable statistician but the softer stuff like this looks like it is outside his comfort zone.
Dec 6, 2011 at 7:51 PM | Unregistered CommenterThinkingScientist
>> It is also important to consider microclimate as well.
It's just occurred to me that this might be rather important. Where do people build cities? Generally not on cold wind-swept plains, but in sheltered valley areas with reasonable water supply - both requirements will moderate the local climate.
And since most weather stations are sited close to cities, they are likely to be equally benign environments. I guess the "experts" probably adjust for altitude, but do they adjust for local wind sheltering, for local rivers, etc?
Urban heating or otherwise is a complex subject.
For example the Weekend Effect where weekend days are significantly warmer than weekdays shows that some urbanization effects cause cooling. In this case urban aerosols is the cause of the Weekend Effect. They decrease at the weekend resulting in more solar insolation and warmer temperatures.
BTW, this explains Mann's and the IPCC's fondness for urban data from China to refute UHI - the high levels of aerosols there that have increased over recent decades producing a cooling trend.
Then there is the Urban Irrigation Effect which cools temperatures due to the greater thermal capacity of water vapor. When they moved the Perth official station from next to an irrigated park to an unirrigate field, average temperatures rose by more than 1C.
Another effect is the huge increase in airconditioning in tropical cities over the last 50 years. Places like Singapore have warmed faster than than the global average in recent decades.
I'm not disputing that UHI is real, just pointing out the picture is rather more complex than many think.
Some of these comments refer to the acknowledged difference in temperature between town and country. I thought we had moved on to understanding that the issue is about whether the global historical record is contaminated by the the growth of urbanisation. If we harp on about towns being warmer than rural areas while Phil Jones says he has accounted for it, then we sceptics can be justly be criticised for not understanding the science.
Lots of interesting comments here; hopefully I can address them all satisfactorily.
For Bruce and Theo, this paragraph in the introduction is of relevance to your concerns:
"An analysis of the impact of urbanization on temperature trends faces multiple confounding factors. For example, an instrument originally installed in a city frequently will have warmer absolute temperatures than one in a nearby rural area (especially at night), but will not necessarily show a higher trend over time unless the environs change in such a way that the UHI signal is altered in the vicinity of the instrument. Similarly, microsite characteristics that may be unrelated to the larger urban environment can have notable effects on temperature trends and act counter to or in concert with the ambient UHI signal."
Put simply, while there might be a 5C or larger average difference in absolute temperatures between urban centers and rural areas, this will not translate to a 5C bias in the trends unless the instrument currently in the urban center started in a pristine rural site and had the city grow up around it. Most urban-located instruments post-date rather than pre-date the urbanity of the location, and its a lot more difficult to infer the effects of changing composition of existing urban areas on the trend (Mosh has a theory that the effect of urbanization on absolute temperature reaches a limit fairly quickly).
David S,
I fail to see how we "persist in using the data in the climate records at face value." The only data we take from the climate records is the raw (unhomogenized) temperatures and the lat/lon coordinates, as we have no alternatives to consider (we actually do compare WMO station lat/lon records to GHCN metadata, but I digress). All of our other metadata for both current conditions and historical changes is taken from independent sources, specifically:
"A complete set of metadata is calculated for each station using the station location information provided in station inventories and publically available GIS datasets. These datasets include: Distance From Coast (0.1 deg), Hyde 3.1 historical population data (5 arc minute), 2000AD Grump Population density (30 arc seconds), Grump Urban Extent, Land use classes from the Harmonized Land Use inventory (5 arc minutes), radiance calibrated Nightlights (30 arc seconds), ISA- Global Impervious Surfaces (30 arc seconds), Modis Landcover classes (15 arc seconds), and distance from the closest airport (30 arc seconds). In addition, area statistics at progressive radii are calculated around each putative site location."
Also, we go to great lengths to look at the sensitivity of our results to the stringency of our urbanity criteria, varying both the distance we look from the station for urban areas (from 5 km to 20 km) and the amount of urban pixels (or nightlights, ISA, pop growth, etc.) that we use as a cutoff. While you start to run into spatial coverage issues at the extremes (e.g. comparing rural stations primarily in the US and AU to all the world's urban stations), the results don't change that dramatically (we included a pair approach to explicitly control for spatial coverage).
Zeke - a year or so ago you did an emulation of the Menne output but were using what was for me an unknown stats package. I tried to get a copy but the old version I found available to download was not able to run your code. Are you using anything different these days such as R?
Clivere,
Are you referring to the station siting analysis here: http://rankexploits.com/musings/2010/a-detailed-look-at-ushcn-mimmax-temps/ ?
Both Mosher and Nick Stokes have packages in R that can do this. Just use Anthony's CRN ratings to select which stations to run through and compare the results. You can do similar things with different urbanity designations (as we do in our poster).
Zeke - thanks - that looks like the one ie CRN12 and CRN345 Stations. Got a url? Also do they have any code that will run with the earlier ushcn file format?
I dont know R but I downloaded it and will give it a go if I have a starter example to work off.